This was presented by Chioma Agwuegbo from Reboot at the Impacts of Civic Technology Conference (TICTeC 2018) in Lisbon on 18th April 2018. You can find out more information about the conference here: http://tictec.mysociety.org/2018
2. Nigeria is one of the most diverse countries, with over 500 ethnic
groups, a mixed of adherents of Christianity, Islam and
traditional religions and a vast geographical peculiarity that
shapes variegated cultural presence
Although these diversities could be strengths, in Nigeria, they
have tended to generate a fractious politics with an unhealthy
competition, generating conflicts which are often violent
One of the major conflicts was the 3-year civil war that was
fought between 1967-1970 in which thousands were killed
However there have been many communal conflicts in virtually
all parts of the country.
Of recent two most serious have been the Boko Haram
insurgency in the northeast part of the country and the farmer-
herdsmen conflicts that has engulfed both north west and north
central parts of the country
3. Many violent conflicts in Nigeria are triggered by hate speech. This is
particularly so in the context of election related violence
As we progressed towards the 2015 general election, we realized that
the increasing deployment of hate speech could lead to violence and so
decided to engage in countering hate speech in order to contain its
potentials to catalyze violence
We also came to the realization however, that countering can only be
effective if we have clear understanding of the sources of hate speech,
the type hate prevalent and the channels through which hate speech was
propagated
Consideration of these led us to mount a hate speech monitoring
platform , enabling us to monitoring and analysis hate speech in the
social media in Nigeria
Building on the experience of iHub Kenya’s Umati project, CITAD
mounted a modified version to monitor and counter hate speech during
the electioneering campaign in 2015 and later added additional features
to monitor the post-election period
More recently, functionality was added to deal with the propagation of
fake images in the context of new conflicts which have implications for
the 2019 general elections
4. We define hate speech as any speech act that has
reasonable chance of catalyzing or amplifying
violence by one group against another, given the
circumstances in which it was made or
disseminated
A Speech act in this context includes any form of
expression, including images such as drawings
or photographs, films, etc
In our categorization of hate speech, we drew a
lot from the work of Susan Benesch who has
done a lot of work in area in both the USA and
Africa
5. Susan Benesch schema has so far remain the more
widely used in identifying dangerous speech. She
specifies five elements namely:
A powerful speaker with a high degree of influence over the
audience
An audience that has grievances and fear that the speaker can
cultivate
A speech act that is clearly understood as a call to violence
A social or historical context that is propitious for violence, for
any of a variety of reasons, including longstanding competition
between groups for resources, lack of efforts to solve grievances,
or previous episodes of violence
A means of dissemination that is influential in itself, for example
because it is the sole or primary source of news for the relevant
audience
The schema does not require that the five elements
must be present for the act to be dangerous
6. Insults people for their religion
Abuses people for their ethnic or linguistic affiliation
Expresses contempt against people because of their place of
origin
Disparages or intimidates women or girls because of their
gender
Condones discriminatory assertions against people living with
disability
Abuses or desecrates symbols of cultural or religious practices
Denigrates or otherwise ridicules traditional or cultural
institutions of other people
Deliberate spread falsehood or rumours that demeans or maligns
or otherwise ostracizes other people on the basis of religion,
ethnicity, gender or place of origin for the accident of one form
of disability or the other
7. We deployed a slightly modified version of the
system that i-Hub used in Kenya for the Umati
Hate speech project
This required the used of human monitors to
browse through the social media using a
generated hate speech lexicon and source list of
suspected hate speech profiles and platforms
In the post election period, we modified this
deploying a semi-automated system (based on
lexicon) and manual system(based on source list)
It also allowed for crowd-sourcing of reporting
by the public but which is then tested to see to
see if reports met the criteria for hate speech
8. •The exercise captured over 680 hate speech contents with
32 attributes of each entry making the total sum of about
21,760 entries, gathered by the 10 monitors over a period
of 5 months ( January to April, 2015)
•In the year 2016, we captured over 6000 hate speech items
•In the year 2017, the figure jumped o over 8000 items
•Where us in 2015, we had only two categories (election
related and others), the subsequent years we expanded to
eight categories to enable us observed more closely the
dynamics of the generation of hate speech in the country
•Each speech item was captured along with its attributes as
entries pre designed in the coding sheet such as: date of
entry, gender of the speaker, Link to the speech item,
language the audience being addressed in, text of the
9. Language of dissemination:
•In the year 2015 about 92.69% of hate speech items
recorded were in English , hence English remains the most
commonly used medium of the circulation of hate speech.
•There was a visible number of comments in Hausa which
took 2.5% followed by Yoruba and Pidgin English which took
1.5%and 0.04% respectively
10. Use of Coded Language:
Only 5% of the hate speech items captured used coded
language in addressing the audience, this involves the use
of vernaculars that compare a group of people with
animals(de-humanization), insects or a derogatory term
used in mother tongue or local proverbs which describe
another group as invaders, foreigners, or interlopers.
11. • The call to action indicates what should what the
speaker wants to be done against the target of th
hate speech
• There is significant call for the killing of members
of other (at 19%) while 7.92% called for riot and
4.3% contains call for forceful eviction of members
of a particular group
12. Historical and social Context:
• We examined whether the speech item had been used
and led to violence. About 36% of the speech items were
found to have been used and led to violence
• Use of words and other speech items such as prophetic
cartoons and other anti religious items were found to be
common. Spread of rumors similar to the ones that
caused violence before was found during the period
13. MEDIUM OF DESSEMINATION :
The data revealed that major online medium used for dissemination of
the dangerous speech is Facebook (with about 76% of the Dangerous
speech items recorded during the period either as a post in a private
Page/group ,a Post in a public page/group or a response to a
post/forum.
Use of Online article for dissemination of Dangerous speech was found
to be the second largest medium with 12% of the sighted dangerous
speech followed by Twitter with 9% and picture with 3%
14. 77%
16%
7%
Does the speech item suggests discrimination against women/girls ?
No Not sure Yes
77%
1%
17%
5%
Gender of the audience Most likely to React
Both Female Male Not know
15. Many of the hate speech items are responses or
commends to what was said in the broadcast media
A number of people vent their anger in the social media in
response to their inability to have the right to reply or
response in the broadcast media
Most media outfits, including radio stations have also
online platforms for their news and political programmes
A number of political programmes have fans pages,
created not by the Radio stations but by people who are
listeners (for example Kawanne Gauta on Freedom Radio in
Kano State): these fan pages generate large volumes of
hate speech
Reducing hate speech will therefore require enforcement
of regulatory provisions with respect to inflammatory
speech
16. They are predominantly males, educated
(since they communicate mainly in English)
and presumably young people(presumably
because age falsification is common in the
social media)
Over 60% are identifiable commentators: they
are partisans of the various political parties
Politicians come only second as a category
but as the result shows, audiences tend to be
more receptive to messages by the politicians
17. DATE EVENTS No
1/3/2015 Pastor Mbaka's Anti Jonathan Preach 1
1/14/2015 Koffi Ana Peace Pact 2
1/19/2015 Governor Fayose death advertorial 3
Dokubo Asare's War threat
1/22/2015 1. Sambo Dasuki’s Call for Postp postponement of Election
2/7/2015 1. Jega Announcement for the postponement of Presidential Election
1. President Jonathan Conceconcieve defeat
February 14th to March
28
1. Announcement of Presidential electt election
1. First Lady Dame Petience's A Almajiri Hate speech
18. The figure below shows the trend of hate speech in the period
January to May, 2015. It shows that it was steadily decreasing
following the peace accord and more elevated campaign for
peaceful elections
19. The data shows that there is connection between
offline events and online dangerous speech
A large number of the items are reaction to
criticism of specific persons. While the criticism
itself is not necessarily dangerous speech, the
reaction tend to qualify as dangerous speech:
what this illustrate is the element of intolerance
of criticism in the society
Tackling or combating online dangerous speech
therefore has to take on board more campaign
on getting people to not be provoked
20. Results are shared with the public using social media
At the end of each month, a monthly press conference
is held during which analysis of the result is shared with
the media
A monthly advisory brief is also shared with a large
number of key stakeholders including security agencies,
National Human Rights Commission, etc
The production of policy briefs which highlight the
results, their implications and what should done to
address both drivers and triggers of hate speech in the
country
Appropriate responses to certain hate speech items
were countered using the same channels
21. We have a better understanding of the
nature of hate speech
We have clearer understanding of its
channels of propagation
We have an understanding of the typical
profile of those who engage in hate speech
We are able to identify key moments that
generated hate speech from which we are
able to identify key triggers and drivers of
hate speech
22. The results serve as input to an early warning system-
using the historical analysis, when hate speech which had
been used before and led to violence occur again, it
means we must take immediate actions to preempt the
possibility of violence
They allow us to develop more effective strategies for
countering hate speech in the country
They allow us to understand better the drivers and
triggers of hate speech in society. This can be seen
through an analysis of the flow of hate speech and events
of offline
They are being used to raise awareness and to drive
advocacy for both government to develop appropriate
responses to drivers of hate speech and for citizens to
not only not engaged in hate speech but also not be
provoked by it
23. There is increasing awareness in the country about the
dangers of hate speech
Results were fed to law enforcement agencies such as the
police and in some instances they have used them to preempt
crisis
A number of agencies have now responded to address issue
of hate speech. For example, the National Broadcasting
Commission has as result undertaken a review of the
Broadcasting Code to incorporate provisions on hate speech.
Similarly, the National Orientation Agency has rolled out a
national campaign against hate speech
There are now a number of bills in the national assembly
aimed at curbing hate speech
There is more a multiplicity of voices working against hate
speech unlike when we started with ours was a lone voice